,__._ __- “I

CONSONANT-TONE INTERFERENCE IN CHADIC

AND ITS IMPLICATIONS FOR A THEORY OF TONOGENESIS

IN AFROASIATIC

Ekkehard WOLFF

(University of Hamburg)

1. INTRODUCTION : AIMS AND HYPOTHESES

The Chadic languages are spoken in the wider Lake Chad

Basinwest, eastandsouthof Lake Chad. They constituteoneof

the southernmost families of Afroasiatic, together with the

Cushitic and, if one is willing to accept a fairly recent

hypothesis, the Omotic languages. Interestingly, interms of

linguistic typology, only these southernmost families of

Afroasiatic exhibit tonal phenomena. Neither Semitic nor

Berber have ever been suggested to be "tone languages", at

least to my knowledge, and there is little hope to learn

something in this respect from the Egyptologists.

This paper was originally prepared to be discussed atthe16th West African Languages Congress in Yaounde, April 1984. For reasons beyond the control of organizers and participants, the Congress could not be held. The present paper is a considerably enlarged and modified version oftheonereadat the 22nd Deutscher Orient+listentag in Tübingen, March 1983, which was published in A@L&LX und Ubt.~ec~ 66 (1983) under the title "Tonogenese in tschadischen Sprachen".

- 194 -

As for the Chadic and the Cushitic languages (and for

Omotic, as little as we know about those languages), their

"tonal" characteristics appear to be far from uniform. Ra-

ther, as for instance in the case of the languages of the

socalled Sam group (Somali and others, cf. HEINE 19781, the

suprasegmental phenomena in question have also been classi-

fied as "accentuai" rather than tonal, or even as border-

line cases between accent and tone. (For a recent analysis

of Somali as a "tonal accent language" see HYMAN 1981 and

further references therein, among them August KLINGENHEBEN's

approach to the question "1st das Somali eine Tonsprache ?"

of 1949.)

As for the Chadic laquages in particular, which con-

tain the bulk of the Afroasiatic languages still spoken to-

day, they have hitherto been treated implicitly or explici-

tly as tone languages without any further indication as to

specific type. It is perhaps interesting to note that, ear-

lier in this Century, a fine linguist such as August KLIN-

GENHEBEN from the University of Hamburg, who was very expe-

rienced in the field of West African languages, has never

guite gotten round to accepting Hausa as a "tone language"

in the same sense as, for instance, Ewe or Vai, i.e. lan-

guages of the "Sudanic" type, to use a term virulent in

KLINGENHEBEN's days. Far away from denying Hausa the sta-

tus of a tone language, we would still today hesitate to

file it away together with, for instance, Yoruba under the

same typological heading.

Viewed in an inter-branch perspective, the guestionmust

arise as to which historical development we ought to envisa-

ge to account for the neat distribution of the feature "to-

nality" within Afroasiatic :

(a) If we assume that tonality was a feature of the common

ancestor Proto-Afroasiatic (PAA), then we ought to be able

to propose a theory which would not only describe the pro-

perties of the PAA tone system but would also explain why

and how the northernmost families, i.e. Semitic and Berber

- 195 -

(we might never find out about Egyptian), have lost their

tonality feature ; under such a theory, we would simple as-

sume that the southern families, i.e. Chadic, Cushitic, and

possibly Omotic as a separate family, have retained their

tonal characteristics from PAA days and we should then be

able to reconstruct the PAA system from the evidence provi-

ded by the latter.

(b) If we assume that tonality was not a common typological

feature of PAA, we ought to be able to propose another theo-

ry which would have to explain why the southern families

show characteristics of tone languages, i.e. whether or not

this typological feature was acquired by southern AA-langua-

ges from the same source jointly or independently through

diffusion (language contact) or by convergent development

independent of each other and of contact with non-AA tone

languages.

Since the Afroasiatic scope would be too wide for a

short conference paper, 1 shall restrict myself to the ques-

tion of "How did Chadic languages acquire tone - if they

did not inherit it from (pre-)Prote-Chadic ?". This highly

interesting question has hardly ever been seriously pro-

pounded. 1 shall, therefore, not discuss theory (a) above

which assumes that PAA was a tone language. 1 shall discuss

theory (b) under which we assume that PAA was non-tonal,

and that southern AA languages, including Chadic, have some-

how acquired "tonality" in the course of

If we pursue this theory further, we

alternative :

(a) Chadic languages have acquired their

external influence, i.e. by interference

languages (diffusion, language contact).

(a.11 If this had happened in fairly

their history.

face the following

tonality through

from other, non-AA

early times, i.e.

before Chadic split into various branches, the tonal sys-

tems of a11 modern Chadic languages ought to be traceable

to this common source. We would then simply be forced to

accept the hypothesis that Proto-Chadic became tonal when

- 196 -

larger numbers of speakers of non-AA tone languages began

to shift to the use of pre-Proto-Chadic dialects, thereby

transferring prosodie features of their original languages

to the new language ("substratum hypothesis" ; this would

have been, basically, the idea of the followers of the "Ha-

mite theory" earlier in this Century).

(a.21 If, however, such language contacts occured in

later times when Chadic had already split into its various

branches, subgroups belonging to different branches but with

a significant areal distribution ought to share typological-

ly similar tone systems. This would again point to a common

"substratum" albeit more limited in distribution and more

recent in relative chronology, i.e. we might wish to accept

the hypothesis that in a particular section of the geogra-

phical distribution of Chadic languages these came into

"contact" with non-AA tone languages of a particular type,

with heavy bilingualism resulting on the side of the origi-

nal non-Chadic speakers who eventually outnumbered the ori-

ginal Chadic speakers and thus established their original

prosodie system as part of the horizontal transmission of

these Chadic languages. A look at the map further below

which shows the areal distribution of a particular type of

prosodie system within Chadic would lend support to this

hypothesis.

(b) Chadic languages did not acguire tonality through con-

tact with non-AA tone languages, but developed tonality in-

dependently, albeit triggered off by the same underlying

universal properties of the human language which has led to

tonogenesis elsewhere, and not only in Africa.

Of course, a third possibility would be that some Chadic

languages underwent development (a), others underwent lb),

and others again underwent both (a) and (b) in the course

of their history. This third possibility shall not be dis-

cussed any further in this paper.

The simplest theory is the one sketched out under (a)

above, which 1 shall henceforth refertoasthe "interference

- 197 -

model". This theory is probably the one which has attracted

the most followers : it is, as has been pointed out already,

related to the notorious "Hamite theory" and is probably as

wrong. Nevertheless, it is propagated by one of the most in-

fluential Chadicists of our times, i.e. Herrmann JUNGRAITH-

MAYR (1979, 1980). JUNGRAITHMAYR, in his writings, has never

been interested in the more general question of how and why

Chadic languages became tonal ; he is entirely preoccupied

with only one minute aspect of the question, i.e. why tonal

marking ("abton") seems to suddenly occur in the Chadic ver-

bal aspect system at the expense of the more Semitic-like

"ablaut" system (which is implicitly assumed to be non-

tonal). The "interference model" and the statements on the

theoretical issues behind it by its adherents are characte-

rized by extreme vagueness of conceptualization and verbali-

zation, cf. for instance, the following quotation from

JUNGRAITHMAYR (1980 : 79) which, because of that, is almost

untranslatable and is therefore given here in the German

original (expressions of vagueness underlined by me) :

"M8gLkh~tie gibt es auch innerhalb des Nigritischen die Abfolge der Stufen III und IV, wie sie oben für das Tschadische festgestellt worden sind : Der Typus der Stufe-III-Sprachen bande sich in einem solchen Fall vor allem in den Adamawa-Ubangi-Sprachen, der der Stufen IV in den Sar (a)-Sprachen bewahrt. Da aber eine gleichar- tige Entwicklung in benachbarten Sprachgruppen LCkt vontinda gtinzfich unabhangig aufgetreten sein düt&t~, kkgt d.ie Annahme nahe, dass zuncindtit die Tendenz der Entwicklung von einer Stufe III su einer Stufe IV - zusammen mit dem Faktum des 'Abtons' - von den nigritischen auf die Tschad-Sprachen ilbergegangen ist."

In his 1979 article on the subject, being the publication

of a paper which was heavily criticized when it was origi-

nally presented to the 14th West African Languages Congress

in Ife, Nigeria, in 1976, the theoretical issues involved

are reduced to the following meagre statement :

"It is, however, probable that these languages in the course of their history and manifold contacts came under the influence of non-Chadic languages whose aspect system may have been similar to that presently found in the Chadic 'abton' languages." (p.137)

Even more refined versions of the "interference model"

would provoke one major objection. If the prosodie features

..< _ ..-_ __

- 198 -

involved, i-e. a particular type of consonant-tone interfe-

rente (sec below), could clearly be established as genetic

property of languages of a particular non-AA languagegroup

likely to have been in "contact" with certain Chadic lan-

guages in the past, one might be forced to accept that mo-

del. If, however, such features cari be shown to be univer-

sa1 properties of the human language and, therefore, cannot

be diagnostic for a particular genetic group of the world's

languages, the theory becomes unacceptable as a mode1 for

explaining the occurence of these features across genetic

language boundaries.

The only other persons, besides JUNGRAITHMAYR and much

earlier, who have sofar addressed their attention to the

problems involved, albeit along a different line of argu-

ment geared towards universal features of tone, were HYMAN

and SCHUH (HYMAN 1973, HYMAN / SCHUH 1974). They, too, did

not quite believe in the "interference model" which appears

to have been implicitly accepted by generations of Africa-

nists :

"It is entirely possible that the Chadic languages owe theirtonal nature to long contact with Niger-Congo languages. However, the examples of tonal processes we shall cite have evolved independent- ly in most cases in Chadic laquages having no present contactwith Niger-Congo languages, and vice versa. These individual examples cannot therefore be directly attributed to contact phenomena. More crucially, the great similarity of detail of these phenomena in Niger-Congo and Chadic languages quite certainly rules out chance convergence." (HYMAN / SCEIUH 1974 : 83)

Note that HYMAN and SCHUH made this statement on a very thin

data basis as far as Chadic languages were concerned : only

three Chadic languages had been surveyed (a11 of them West

Chadic), one of them, Ngizim, is guoted with examples, and

LUKAS' treatment of Bole consonant-tone interference is re-

ferenced in the bibliographical section, together with

NEWMAN's then unpublished Kanakuru grammar.

What 1 wish to establish in this paper is the plausibi-

lity of alternative (b) above, i.e. the likelihood of inde-

pendent development of tonality in Chadic on the basis of

- 199 -

universal properties of the human language. Whether suchde-

velopment was reinforced by longer periods of bilingualism

involving true tone languages, or whether certain typologi-

cal developments must be attributed to such contact situa-

tions, is of only secondary importance here. Far fromdenying

possible interference from Niger-Congo languages, it should

become clear that assumptions about language contact arenot

necessary to explain the emergence of tone in Chadic.

The specific vantage point from which to discuss a plau-

sible theory of independent tonogenesis in Chadic is that of

a particular type of consonant-tone interference, well known

to Bantuists since the 1920s in connection with the notion

of "depressor consonants". Concerning tonal depressors and

the issue of tonogenesis it may be useful to quote an autho-

rity :

"The different effects on tone by voiceless and voiced oral obs- truents are manifested today even in nontonal languages... Dia- chronically this effect has given rise to the introduction or mul- tiplication of tones..." (HOMBERT 1975, guoted in OEIALA 1978 : 26).

Vhe development of contrastive tones on vowels due to the loss of a voicing distinction on obstruents in prevocalic position is pro- bably the most well documented type of tonogenesis. When such de- velopment occurs, a relatively low pitch register develops on vowels following the previously voiced series, and a relatively higher pitch is found after the previously voiceless or aspirated series. This process cari lead to a multiplication by two of the number of tones. If the language os atonal, it Will have two tones after this development ; an already existing two-tone system cari be transformed into a four-tone system, and SO on." (HOMBERT 1978 : 78).

The Chadic evidence for such type of tonogenesis Will appear

to fit and, at the same time, not to fit these widely accep-

ted ideas on the emergence of tone. On the one hand, the

voicing distinction of syllabie-initial obstruents plays the

all-decisive role in assigning pitch realizations to follo-

wing vowels. On the other hand, the voicing distinction is

still phonologically operative in these languages, and there

is no indication whatsoever that the contrast between voiced

and voiceless obstruents, for instance, is being given up.

We do, however, observe historical changes concerning the

- 200 -

feature "voiced/voiceless" (in some instances involving the

feature "glottalized' as well) within individual Chadic lan-

guages or language groups (cf., for instance, NEWMAN 1977 :

15ff. for the West Chadic Angas group and Central Chadic

Tera and Bura groups ; WOLFF 1984, 1985a, 1985d for Central

Chadic Musgu, Zime-Mesme of the Masa group, and East Chadic

Kwang) .

Furthermore, and this makes the Chadic case even more

interesting, some languages appear to have two coexisting

systems which are both built on pitch level distinctions,

i.e. non-distinctive "pitch patterns" on the one hand, and

fully contrastive "tone patterns" on the other, neatly dis-

tributed over different areas of grammar and lexicon. In

historical terms : in certain areas of grammar or lexicon

pre-existing pitch distinctions have been phonologized SO

that we are justified to speak of "tone". At the same time,

in the same language, other areas of grammar or lexicon have

not phonologized the pre-existing pitch distinctions SO that

no grammatical or lexical oppositions are made use of. 1

shall exemplify such "dualistic" pitch/tone system by using

data from Lamang, a Central Chadic language of the western

subbranch (for details of description cf. WOLFF 1983a) :

only two verb bases, but a11 nouns are characterized by ba-

sic, yet non-distinctive pitch structures which are predic-

table from the distribution of syllable-initial consonant

types. Two types of non-syllabic segments are distinguished

according to their effects on the pitch realization of the

following vowel : non-depressor (sonorants, glottalized and

voiceless obstruents) and depressor (voiced obstruents).

Syllables which begin with a depressor consonant are reali-

zed with a lower pitch than syllables which beginwitha non-

depressor. The depressor effect is most clearly seeninword-

initial syllables :

non-depressor depressor

H: t1a "Cow" L: ghwh "mountain"

H-H : tsxura "to sit" L-L : gùrvà "to dance"

- 201 -

On the other hand, lexically distinctive tone occurs with

the word class of the expressives (ideophones, etc.) and

with the socalled grammatical morphemes of the language

(bound morphemes, prepositions, determiners incl. numerals),

e-g.

H: btid expr. "pus coming out of a swelling"

L: bud expr. "people being thrown out of their

settlement"

H-L : :SI verb ext. separative/partitive

L-H : :s: verb ext. increase/addition (frombelow)

H: 6 prep. in, into

L: 3 wep. by, by means of

In addition, tone has a heavy functional load in the verbal

aspect/tense system (grammatical tone), e.g.

L-L : kàli continuous "1 take"

H-H : kali durat. cent. "1 keep taking"

H-H-H : kakali Perfect. 1 "1 have taken"

L-L-H : kakàli Perfect. II "1 have begun to take"

On top of a11 this, Lamang uses both tonal accent with nouns

and morphophonemic stress with a limited set of construc-

tion types.

At one time in their linguistic history, therefore, other

Chadic languages might as well have known the cooccurence of

non-distinctive pitch patterns and distinctive tones. In such

a constellation, it is quite likely that "tone" (most likely

starting off from the prominence of a particularly high pit-

ched syllable) - as opposed to non-distinctive pitch -resul-

ted from a previous accent system. That and how a tonal ac-

cent system developed into a tone system independent of

language contact in a Chadic language has recently been es-

tablished for Podoko, a western Central Chadic language clo-

sely related to Lamang (WOLFF 1985c). 1 have argued there

that Proto-Podoko had a system with grammatically conditio-

ned prosodie specification at least for nouns - very much

like the system recently postulated by HYMAM (1981) for

- 202 -

Cushitic Somali, quite distantly related to the Chadic lan-

guages under.review. In such a system, tonal accent accom-

panies the marking of categories in the inflectional and/or

derivational morphology. TO quote an example

the fact that a noun of present-day Dodoko's

like kada "dog" has the following underlying

representation

/kada/

from Podoko :

tone class 3

suprasegmental

is explained by deriving it from a Proto-Podoko "modified"

stem which carried a suffix beginning with a depressor con-

sonant and therefore carrying a low pitch. Such modified

noun stems obligatorily carried penultimate tonal accent.

This tonal accent placement resulted in the association of

a H tone with the penultimate syllable of the modifiedstem,

the suffix retained its low pitch conditioned by itsinitial

segment :

*/kada-CV/

II I / -1 +H +l /

(The non-distinctive basic pitch levels "10~" and "non-1ow"

are indicated by using small letters, the tonal accent is

indicated by +H.) The segmental part of the suffixwaslater

dropped, grammatical marking became reduced to solely pro-

sodic features, i.e. the unmodified stem would have had the

entirely predictable pitch contour non-low followed by low

(*/h-l/), while the modified Stern, after the loss of the

modifying suffix, had the contour non-low followed by high

(*/h-H/) with a "floating" low pitch/tone following the ac-

centual H tone. Thus, the present day noun for "dog" repre-

sents not the simple but the modified Proto-Podoko noun

stem which was trisyllabic rather than disyllabic, therefore

a pattern of three tones has to be associated with the pre-

sent-day disyllabic reflex : H-Ht-L).

NIGER

WEST-CHAVTC

0 100 200Km 1

'\

TCHAD

EAST-CffAVlC

l GAROUA

I

N 0 W

I

- 204 -

2, THE LINGUISTIC EVIDENCE

For the conspicuous geographical distribution of the at

least 13 Chadic languages in which consonant-tone interfe-

rente has been observed cf. the following map.

For the genetic affiliation of the 13 languages to be

discussed, based on a comparative analysis of segmentalpho-

nology and lexicon (NEWMAN 19771, cf. the following diagram-

me in which the branches and sub-branches are arrangedfrom

left to right in roughly west-east order :

PROTO-CHADIC

WEST-CHADIC CENTRAL-CHADIC MASA GROUP EAST-CHADIC

A B

\ \ Bole, Ngizim,

Kanakuru, Bade,

A B A B

\ I I \

Podoko, ~VJ, Masa, Kera, . . .

Lamang, . . . Zime, Kwang,

Lame, Tobanga,

. . . . . .

In the following sections of the paper, 1 shall present

some illustrative data to show that and how the particular

consonant-tone interferences operate in these 13 languages.

In the sequence of presentation 1 shall begin with the Masa

Group and Central Chadic in the tore area of the geographic

distribution of the features under review before illustra-

ting evidence from the more peripheral groups from East-

and West-Chadic.

2.1. From the Masa Group, Masa itself has been shown to dis-

tinguish three types of syllable-initial consonants : de-

pressors (voiced obstruents), non-depressors (a11 other

- 205 -

consonants with the exception of nasal plosives), and 'neu-

tralH consonants which occur with both Low and non-Low tones

on the following vowel (nasal plosives) (cf. CAÏTUCOLI 1978).

The following verbal nouns show the predictable tones on

their stem syllables :

depressor L ba-na "perdre"

L zik-r)a 'lmonter"

non-depressor M pi-na IIplanter"

M cuk-r)a "semer"

M lut-na "écraser"

neutral L nik-xJa lltomberW

M nus-na 'nagerl

2.2. SACHNINE's little Lame vocabulary (1978) suggests a

similar situation in this Masa Group language as well, al-

though the membership of the "neutral class" may be quite

different, cf. for instance examples with initial /h/ in

addition to nasal plosives and prenasalized obstruents :

/h/ L hùtù "grind"

H hum "car"

/m/ M mata "corpse" to be compared with

H mata "die"

/n/ L nè "g ive "

H naw "cow" (loan from Fulfulde ?)

/nd/ L ndè "fall"

H ndarwa "scorpion"

(Note in this context that in East Chadic Kera /h/ is the

only neutral consonant.)

From the wordlist it Will appear that Word-initial voi-

ced obstruents are regularly followed by Low tone, voiceless

and glottalized obstruents are followed by High tone, al-

though the details of Lame phonology remain yet to be worked

out, especially as regards the underlying nature of the fre-

quent Mid tones.

- 206 -

2.3. The three types of consonants show up clearly in the

Zime-Mesme cluster of languages (JUMGRAITHMAYR 1978 ;WOLFF

1985a, 198533). Cf. the tones of the Mesme verbs in the con-

tinuous construction :

depressor L num gb ggom "he is learning"

non-depressor H num go tUm "he is beating"

neutral L num go hèn "he is filling"

H ntim go hén "he is shaking".

2.4. The only East Chadic language that 1 know of for which

consonant-tone interference has been explicitly described

is Kera (EBERT 1979). Voiced obstruents function as depres-

sors, the only "neutral" consonant is /h/, a11 others func-

tion as non-depressors. Cf. the tones on the first syllables

of the "base" forms of the following verbs :

depressor L gùsi "buy"

L bèl?? "love"

non-depressor H peté "pluck (feathers)"

H mante "call"

neutral L h&fé "Cut"

H haté "learn"

Cf. also the voiced/voiceless distinction of the simple as

opposed to the plural verb bases which is paralleled by a

Low/High distinction :

simple base (L) plural stem (H-H)

dar- tar- > téré "anlegen"

gàr- kar- > kéré "pflanzen"

jam- cam- > cerné "schneiden"

etc.

2.5. With only very few verbs, East Chadic Tobanga has re-

tained traces of consonant-tone interference within the

grammatical subsystem of verb pluralization ; cf. the fol-

lowing exhaustive list (CAPRILE 1978) :

- 207 -

non-plural plural

d&ë (4-4) torë (2-2)

"enlever (un objet)" / "enlever (plusieurs objets)"

dogë (4-4) toge (2-3)

"lancer (un couteau)" / '1 ancer (plusieurs couteaux)'

jIbë (2-4) cibë (2-2)

"lancer (une sagaie)" / "lancer (plusieurs sagaies)"

The integers l-4 are being used to indicate highest to

lowest pitch level. These very few examples nevertheless

clearly indicate that changes from voiced to voicelessini-

tial obstruent are clearly accompanied by increase in pitch

on at least one syllable of the word (cf. Kera above and

Kwang below).

2.6. East Chadic Kwang with its two dialects Mobu and Ngam

which have recently been described (LENSSEN 1984), knows

plural verb stems which are formed along the same lines as

verbal plurals in Kera and Tobanga, i.e. formations that

look like synchronie devoicing of the first consonant of

the verbal base. The identification of the underlying tones

of Kwang verbs is a highly complex matter, certainly

LENSSEN's surface structure descriptions do not give the

true story (the details of a reanalysis of LENSSEN'S account

shall be presented elsewhere, cf. WOLFF 1985d). Suffice it

to show here that the shifting from voiced to voiceless ob-

struent in initial position is accompanied by a shift from

Low to Mid tone on the first syllable of the verb Stern, for

instance in the Ngam dialect. (Note that the plural verb

stems in this dialect appear to be lexicalized ; historical

analysis, however, allows them to be analyzed as original

sg./pl. pairs.)

old simple stem : L-M old plural stem : (M-)H

gèsi "compter" kesi "enterrer"

bàti "frapper" pati "cueillir"

bèâi "incuber" peâi "tresser"

. .I.

consonant-tone interference. There are also other reflexes

of this interference in certain correspondences between seg-

mental structure and tone class membership (capital "C" =

depressor, “c” = non-depressor) :

tone class 2 (a) &Cv (b) &Cv (a) /vaa"a/ "night"

n A (b) /vaga/ "place" HL HL

tone class 3 (a) ctlrcv (b) cv&

I \ HL HL 0 0

tone class 4 (a) cv& (b) cv&

H H

tone class 5 C&G

(a) /cfafa/ "fufu"

(b) /kada/ "dog"

(a) /kata/ "sifter"

(b) /Ytsaga/ "cooking pot"

/buti/ IIa sauce"

LH

The H tone in the underlying representations of Podoko di-

syllabic nouns reflects a morphologically conditioned tonal

accent of Proto-Podoko (WOLFF 1985c).

2.10. West Chadic Ngizim distinguishes again three types of

consonants according to their effect on suprasegmental

structure :

II . I . voiceless obstruents block the spreading of L and voiced obs-

truents block the spreading of H. On the other hand, sonorants exert no blocking effect, but rather allow any tone to spread through them." (HYMAN 1973 : 165)

For instance, a Ngizim H tone is lowered to L when it is

preceded by a L tone and is followed by another H ton and

the syllable begins with either a voiced obstruent or a

sonorant (i.e. a phonetically voiced consonant), i.e.

LHH > LLH :

/mùgba +bai/ > mugba bai llit's not a monitor"

/maarSm+tSn/ > màaràm tan "big nose"

but /gii.ta +bai/ > giita bai "it's not pepper"

--.- l_..,._

- 210 -

A H tone spreads into a following L tone syllable only if

the intervening consonant is a sonorant or a voiceless obs-

truent, i.e. not a depressor :

/na baka tlhwai/ > na bak5 tld!w&i "1 roasted the meat"

(! indicates downstepping of the syllable to its right.)

In the subjunctive verb Stern, the tone of CV verbs patterns

with the type of the initial consonant : H tone if it is a

voiceless obstruent, L tone otherwise, e.g.

H tone L tone

Ci "est" mi "take"

shi "drink"

2.11. The examples last quoted from Ngizim resemble the si-

tuation in Bade. In this language, the socalled subjunctive

stem of the verb which is, possibly, historically the un-

marked unit within its verb system, supports the depressor

theory :

II f . . verbs beginning in voiced obstruents are low, others high."

(scHuH 1977 : 155)

2.12. In West Chadic Bole LUKAS (1969) had already found it

necessary to distinguish between "tone permeable" and "tone

impermeable" consonants. The class of voiced (non-nasalized)

obstruents block the spreading of a preceding H tone onto

the syllable which begins with one of these consonants. The

tone impermeable consonants correspond to the depressor con-

sonants in other languages, the tone permeable consonants

correspond to the non-depressors (in Bole they include so-

norants, glottalized obstruents, voiceless obstruents, na-

salized obstruents). Cf. the blocking of the spreading of

the H tone of the first syllable into the second syllable

in the following examples, once the second syllable begins

with a voiced obstruent :

kab&lttiwooyii < baltu "you have enclosed"

kaktimbawooyii < kùmaa "you have heard"

- 209 -

consonant-tone interference. There are also other reflexes

of this interference in certain correspondences between seg-

mental structure and tone class membership (capital "C" =

depressor, mc” = non-depressor) :

tone class 2 (a) C&v (b) &Cv

n A HL HL

tone class 3 (a) C&v (b) cv&

I I HL HL 0 0

tone class 4 (a) cv& (b) cv&

I I H H

tone class 5 C&G

I I LH

(a) /vada/ "night"

(b) /vaga/ "place"

(a) /cfafa/ "fufu"

(b) /kada/ "dog"

(a) /kata/ "sifter"

(b) /Ytsaga/ "cooking pot"

/buti/ 'a sauce"

The H tone in the underlying representations of Podoko di-

syllabic nouns reflects a morphologically conditioned tonal

accent of Proto-Podoko (WOLFF 1985c).

2.10. West Chadic Ngizim distinguishes again three types of

consonants according to their effect on suprasegmental

structure :

For instance, a Ngizim H tone is lowered to L when it is

11 . . . voiceless obstruents block the spreading of L and voiced obs-

truents block the spreading of H. on the other hand, sonorants exert no blocking effect, but rather allow any tone to spread through them." (HYMAN 1973 : 165)

preceded by a L tone and is followed by another H ton and

the syllable begins with either a voiced obstruent or a

sonorant (i.e. a phonetically voiced consonant), i.e.

LHH > LLH :

/mùgba +bai/ > mùgba bai llit's not a monitor"

/maarSm+tan/ > m&aràm tan "big nose"

but /giita+bai/ > S?ita bai "it's not pepper"

- 210 -

A H tone spreads into a following L tone syllable only

the intervening consonant is a sonorant or a voiceless

truent, i.e. not a depressor :

/na bàk5 tlùwai/ > na bak& tlti!wai "1 roasted the

if

obs-

meat"

(! indicates downstepping of the syllable to its right.)

In the subjunctive verb Stern, the tone of CV verbs patterns

with the type of the initial consonant : H tone if it is a

voiceless obstruent, L tone otherwise, e.g.

H tone

Ci "est"

shi "drink"

L tone

mi "take"

2.11. The examples last quoted from Ngizim resemble the si-

tuation in Bade. In this language, the socalled subjunctive

stem of the verb which is, possibly, historically the un-

marked unit within its verb system, supports the depressor

theory :

II . . . verbs beginning in voiced obstruents are low, others high." (scEmI 1977 : 155)

2.12. In West Chadic Bole LUKAS (1969) had already found it

necessary to distinguish between "tone permeable" and "tone

impermeable" consonants. The class of voiced (non-nasalized)

obstruents block the spreading of a preceding H tone onto

the syllable which begins with one of these consonants. The

tone impermeable consonants correspond to the depressor con-

sonants in other languages, the tone permeable consonants

correspond to the non-depressors (in Bole they include so-

norants, glottalized obstruents, voiceless obstruents, na-

salized obstruents). Cf. the blocking of the spreading of

the H tone of the first syllable into the second syllable

in the following examples, once the second syllable begins

with a voiced obstruent :

kabaltuwooyii < bàltti "you have enclosed"

kakumaawooyii < khmaa "you have heard"

- 211 -

’ fl zoggé < zo3Jgé "to the hyena"

'fi tUrùm < tùrùm "to the lion"

2.13. West Chadic Kanakuru has predictable tones in its

verb system :

"For many verbs tbe underlying tone is almost entirely predictable on tbe basis of tbe segmentals used. If a verb begins witb a voi- ced stop, its tone Will be Hi . ..Lo. Tbere are no exceptions. If it begins with a lùkhre "disperse"

*lapèrè > laper& "hold down"

*wUpe > wi!lpe "sell"

*Wùpè > wiIpè "knot"

3, CONCLUSION

Once we start looking at the linguistic map of the Lake

Chad Basin we are struck by the areal clustering of the Cha-

dit languages mentioned above which a11 share this particu-

lar type of consonant-tone interference : it is the eastern-

most West Chadic, the westernmost East Chadic languages and

the languages between these two groups, i.e. from the Masa

- 212 -

group and from the two subbranches of Central Chadic. This

distribution, whatever conclusions we draw from it, is high-

ly significant.

If the particular feature of consonant-tone interference

in our case is not to be attributed to universal properties

of the human language, the only plausible explanation would

appear to be - under our theory concerning atonal Proto-

Afroasiatic and Proto-Chadic ancestorship - a commom adstra-

tum, most likely of Niger-Congo provenance. This hypothesis

would account for type and areal distribution of the supra-

segmental system in the languages enumerated in this paper-

it would not, however, account for the fact that a good hun-

dred or more Chadic languages are also tonal, yet of diffe-

rent typological class membership - as it Will appear at the

time being.

If, however, and this is what 1 am inclined to think,we

are dealing with truly universal and basic linguistic pro-

perties, then the languages under review might have simply

retained typological features from the earliest times of

their linguistic history and the areal distribution of these

features would be rather coincidental - if it is not an in-

dication that only in the "tore area" of Chadic, independent

of branch membership, these prosodie "archaisms" have been

retained while they have been lost in a11 those subbranches

and groups which have geographically moved away from this

"tore area".

The "retention model" as opposed to the "interference

model" would also account for the observation that conside-

rable differences exist between these languages which cari

hardly be accounted for by a "common adstratum hypothesis",

as to

(i) which consonants, other than the depressors, are

grouped as non-depressors or neutral consonants ;

(ii) which classes of lexemes are affected by consonant-

tone interference : only verbs in one particular inflectio-

na1 category, a11 verbs, only some or a11 the nouns, both

- 213 -

nouns and verbs, etc.

If we are right in accepting the retention mode1 as the li-

kely historical explanation, this would have immediate re-

percussions on other central issues in comparative Chadic

and comparative Afroasiatic research which, to an extent

which 1 find hard to accept, tend to rely on the identifi-

cation of socalled archaisms and, to an extent which 1 find

hard to accept, the labelling of whole languages or groups

of languages as more or less "archaic" or "innovative" (cf.,

for instance, JUMGRAITHMAYR 1977 and ZABORSKI 1983,whoseem

to simply "know" that Central Chadic languages are "rather

innovative", while their "evidence" for such evaluation is

based almost entirely on aprioristic reasoning and is, the-

refore, not at a11 convincing !).

The theory which 1 wish to tentatively propose assumes

that, if or although Proto-Chadic was atonal, different

pitch realizations, based on the nature of syllable-initial

segments, played already a role in the earliest periods of

Chadic linguistic history, most likely in combination with

"tonal accent" phenomena. And that it was part of the indi-

vidual language's history to develop different ways to pho-

nologize and grammaticalize and/or lexicalize the inherited

universalistic pitch realizations (quite likely under the

influence of tonal accent phenomena ; cf. WOLFF 1985c, 19858).

In short : for Chadic languages to develop tonality, it is

not necessary to postulate interference from non-Chadic lan-

guages. Rather, several languages in the tore area of Chadic

language distribution which belong to a11 four branches pre-

sently distinguished, suggest that independent tonogenesis

is at least as likely a theory to account for the facts as

is any other - if not a likelier one.

As for the wider application of the results of compara-

tive Chadic suprasegmental phonology to Afroasiatic linguis-

tics, Cushitists and Omoticists should feel encouraged to

apply the method of interna1 reconstruction to the supra-

segmental structure of their languages. It is quite possi-

ble that they Will corne up with reconstructions of Proto-

- 214 -

Cushitic and/or Proto-Omotic tonal accent systems which are

typologically similar to what we have begun to discover in

Proto-Chadic. It would be rewarding to soon read a paper on

tonogenesis hypotheses for Cushitic and/or Omotic. From the

Semitists one would like to read about a comparative study

on accent patterns from the earliest sources to the various

dialects of Semitic languages spoken today, to be comple-

mented by a comparative study on accent in Berber. 1 dare

predict that comparative Afroasiatic suprasegmental phono-

logy Will provide the solutions to more problems of compa-

rative phonology and morphology than we are able to imagine

today.

REFERENCES

ANDERSON S.C., J. SWACKHAMER - 1981 - "From consonants to downstep in Podoko" - Studiti Ln &tican tingtiticb 12 - pp.131-153.

CAÏTUCOLI C. - 1978 - "Schemes tonals et morphologie du verbe en masa" - Phéa&b&n à &x heconbathu~on du photo-&had.iqUe (J.-P. Caprile et A. Jungraithmayr ed.) - Paris : SELAF - pp.67-88.

CAPRILE J.-P. - 1978 - "Notes linguistiques sur le tobanga a partir d'un conte en cette langue" - Cinq texa% titchadiquw (Camvtoun ti Tchad) (H. Jungraithmayr et J.-P. Caprile éd.) - Berlin : D. Reimer - Marburger Studien zur Afrika-undAsienkunde - pp.121-175.

EBERT K.H. - 1979 - Sphache und Th&iti_on deh Kerra (Tbchadl, 3. Gtam- mtik - Berlin : D. Reimer - Marburger Studien sur Afrika- und Asienkunde.

HEINE B. - 1978 - "The Sam languages : A history of Rendille, Boni and Sonkali" - A@wana tingu..L4X.& 6 (2) - mlibu.

HOMBERT J.-M. - 1975 - Towahds a tiwhg 04 Zonogenesti : An emptical, phghiologica&?g, and pe&xp,tuaUg-bahed account 05 Xhe. dewlopment 06 ton& comt in &nguagt? - Unpubl. Ph.D. dissertation - Berkeley : Univ. of California.

- 1978 - "Consonant types, vowel guality, and tone" - Tone : A Lin- cp&~C bUhWg (ed. by V.A. Fromkin) - New York - pp.77-111.

HYMAN L.M. - 1973 - "The role of consonant types in natural tonal assi- milations" - Cotionant tgpk2 and tine (ed. by L.M. Hyman) - Sou- thern California Occasional Papers in Linguistics 1 - pp.151-179.

- 215 -

HYMAN L.M. - 1981 - "Tonal accent in Somali" - Studiti in Afjhican Lin- gui&.&4 12 - pp.169-203.

HYMAN L.M., R.G. SCHUH - 1974 - "Universals of tone rules : Evidence from West Africa" - Lingu.&;tic Znqu.&y 5 - pp.81-115.

JUNGRAITHMAYRH. - 1977 - "Sprachhistorische Schichtstufen im Tschad- raum" - Qaidema 23 - pp.95-100.

- 1978 - "The Zime dialect cluster ("Kado", "Dar:") in Southern Chad : its verbal aspect system" - A&&I und Ubmee 61 - pp.l-27.

- 1979 - "Apophony and tone in the Afroasiatic/Niger-Congo frontier area" - Etida .&ngtiXiqueb (Niamey) 1 (1) - pp.130-140.

- 1980 - "Kontakte zwischen Adamawa-Ubangi- und Tschad-Sprachen : Hur übertragung grammatischer Systeme" - Z&chtr-i@ dm Uetichen Mohgen&ïncl&hen Gc%.W&adt 130 (1) - pp.70-85.

KLINGHNHEBEN A. - 1949 - "1st das Somali eine Tonsprache ?" - Zeti- bch&i& @ih Qhonetih und Al.tgem&ke SphmhUiAbenbcha~~ 3 - pp.289- 303.

LENSSHN T. - 1984 - Stidien zum Vmb im Kwang (Tbchadl : Phonologie und Mo/rphologie - Africana Marburgensia, Special Issue 8.

LUKAS J. - 1969 - "Tonpermeable und tonimpermeable Konsonantem im Holanci (Nordnigerien)" - EXhnoLogical and Lingtitic Stidies in ffonowr 06 N.J. van Wwun&o - Pretoria - ~~-133-138.

NEWMAN P. - 1974 - The. Kanakwtu Language, - Leeds.

- 1977 - "Chadic classification and reconstructions" - A&oUb&c tingti~cs 5 (1) -Malibu.

OHALA J.J. - 1978 - "Production of tone" - Tone : A k?.ing&5fic bUJWLy (ed. by V.A. Fromkin) - New York - pp.5-39.

SACHNINE M. - 1978 - "Liste lexicale en lame (vùn dz*aw) du Cameroun" - Qh&x.kb~~ à Ru hecov&Auc;tion du pk~~~-Xchadique (J.-P. Caprile et I-l. Jungraithmayr ed.) - Paris : SELAF - pp.195-201.

SCHUH R.G. - 1977 - "West Chadic verb classes" - Qapm &I Chadk Lin- gui&%33 (ed. by P. Newman and R.M. Newman) - Leiden - pp.143-166.

TOURNEUX H. - 1978 - Le rrut!wi ou vtim de Mog/roum (Tchad) : Qhonoto- . g.ie, EkZrnenX~ de gmmrna.& - Paris : SELAF.

WOLFF E. - 1983a - A gmmmm 04 ;the Lamang Canguage (C&d Làmùnl - Glückstadt.

A&..ika und übmee - 1983b - "Tonogenese in tschadischen Sprachen" - 66 - pp.203-220.

.__ . -_.,_

- 216 -

WOLFF E. - 1984 - "Adverb and verbal noun formation in the Musgu lan-

guage of Mogroum (Vulum / Mulwi). Studies on consonant-tone inter- ference in Chadic laquages II" - A&h und Ubmee 67 - pp.l75- 197.

- 1985a - "The verbal aspect system in Zime-Mesme. Studies on conso- nant-tone interference in Chadic languages III" - A&&I und übmee 68 (in press).

- 1985b - "Consonant-tone interference and current theories on verbal aspect systems in Chadic" - kYceedingb 06 the hhA% Itiwonti

COngUAb on thn&f_Ob~ f.hIgUkb~tb - Marburg (19831 (in press).

- 1985c - "From accent to tone in Podoko : Tonogenesis and the his- torical explanation of downstep in a Chadic language of northern Cameroon" (forthcoming) .

- 1985d - "An alternative solution to "tone reversal" in East Chadic Kwang" (unpubl. man.

WOLFF E., J. SWACKBAMER - 1985 - "An autosegmental approach to tone in Podoko" (forthcoming) .

ZABORSKI A. - 1983 - "Chadic stratification" - SX~dieb in Chadi~ and A@~oaha.-tic LingbiAti (ed. by E. WLFF and Ii. MEYER-BABLBURG) - Hamburg - pp.233-246.